e journal of remote sensing & gis - longdom · remote sensing by satellites and their study...

Lithological Discrimination by Selective Image Processing TechniquesRaghu Babu K1, Keshava Kiran Kumar PL1 and Ramakrishna P2 *

1Department of Geology, Yogi Vemana University, Kadapa, Andhra Pradesh, India2Government Degree College, Nalgonda, Telangana, India*Corresponding author: P Ramakrishna, Government Degree College, Nalgonda, Telangana, India, Tel: 9440673734; E-mail: [email protected]

Rec date: November 01, 2017; Acc date: December 11, 2017; Pub date: December 13, 2017Copyright: © 2017 Raghu BK, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Abstract

Exploration geology involves mapping of rocks that includes delineation of different rock types known aslithological maps. Such maps provide primary information about the mineralogical and hydrological occurrencesbeneath the surface of the earth. Remote Sensing by satellites and their study through GIS are sophisticatedtechniques available today for lithological discrimination and mineralogical and hydrological exploration. Spatialtechnology through satellite data has become an excellent technology not only for lithological mapping, butlineament extraction, structural mapping etc. In the present study image processing techniques have been applied toIRS P6 LISS III image of 57j/7 scene covering Vempelli, Vemula and Velpula villages of 720 sq.km area. It iscomposed of all three varieties of rock types viz., igneous, metamorphic and sedimentary rocks. Peninsular GneissicComplex (PGC) is the basement rock consisting of granites, granodiorites, schists and gneisses forms the basementof the area, over which Cuddapah Supergroup rocks are present. Gulcheru quartzites, Vempalli dolomites, ofPapaghni group, Pulivendla quartzites, Tadipatri shales of Chitravati group are the Cuddapah Supergroup rocksoccurring in the study area. Igneous activity is seen between Vempalli dolomites, Pulivendla quartzites and withinTadipatri shales in the form of sills.

Keywords: Lithology; Exploration; Mapping; Quartzites; Limestones;Dolomite; PGC

IntroductionImage Processing Techniques key elements in the interpretation of

satellite image for various studies like land use land cover, mineralexploration, groundwater exploration, lithological discrimination,forest cover studies etc. Satellite image of 57j/7 (Figure 1) has beentaken for lithological discrimination by the application of selectiveimage processing technique. The advanced software ERDAS Imagineprovides various image processing techniques such as radiometriccorrection, filtering, classification and PCA techniques. In the presentstudy filtering techniques like High pass and Low pass in 3 × 3, 5 × 5and 7 × 7 and PCA analysis technique are applied for delineation oflithological units of the study area. These maps provide primaryinformation about the lithological distribution over the surface of theearth which is used to infer mineralogical and hydrologicaldistribution beneath the surface of the earth [1]. The study area existsbetween 14°15`N, 14°30`N latitudes and 78°00`E, 78°15`E longitudes.It consists of all types of rocks viz., igneous, sedimentary andmetamorphic rocks. Granites and granioids, schists and gneisses arethe igneous and metamorphic rocks belonging to the Archaeanbasement known as Peninsular Gneissic Complex (PGC) which isseparated from the overlying Cuddapah Supergroup meta sedimentsby an unconformity. Sedimentary rocks comprises of quartzites ofGulcheru formation, dolomites, shales of Vempalli formation ofPapaghni group, quartzites of Pulivendla formation of Chitravati groupand shales of Tadpatri formation of lower Cuddapah Supergroup ofrocks. Igneous intrusions occur in the form of sills [2]. Present studyreveals the impression of these rocks in satellite image and propertiesexerted through various image processing techniques.

Figure 1: Satellite image of the Study area 57j/7.

The main objective of the present study is to differentiate variousrock types of the study area. In general basement granites and

Jour

nal o

f Remote Sensing

& GIS

ISSN: 2469-4134 Journal of Remote Sensing & GISRaghu et al., J Remote Sensing & GIS 2017, 6:4

DOI: 10.4172/2469-4134.1000221

Research Article Open Access

J Remote Sensing & GIS, an open access journalISSN: 2469-4134

Volume 6 • Issue 4 • 1000221

mailto:[email protected]

ganodiorites are rich in quartz and feldspars, hence they show lightertonte and rough texture, qurtzites which are ferrugenous show darkertone. Calcium rich limestones and dolomites show lighter tone andsmooth texture. Shales show lighter tone and smooth texture, Basalticsill bodies occurring in the study area rich in magnesium and iron areshowing dark tone and rough texture and enriched in vegetation.When the image is subjected to digital image processing techniques,these rocks show varied diagnostic properties.

Methodology and MaterialsGeoreferenced IRS P6 LISS III image of Vempalli, Vemula, Velpula

area corresponding to 57j/7 at 1:50,000 scale is used in the presentstudy. Primarily all the rock types of study area are delineated usingGeographic Information System (GIS) 9.1 software and thematic mapof geology of the area has been prepared (Figure 2). Image processingtechniques of ERADAS Imagine 9.1 software such as histogramequalization, 3 × 3, 5 × 5, 7 × 7 Lowpass filtering, 3 × 3, 5 × 5, 7 × 7High pass filtering and Principal Component Analysis (PCA)techniques are applied to study the characteristics of different rockstype of the study area during application of different techniques.

Figure 2: Thematic map showing the Geology of the Study area.

Principal Component Analysis (PCA)Images generated by digital data from various wavelength bands

often results an image difficult to interpret due to redundancy ofmultispectral band data. Principal Component Analysis is the

technique designed to reduce such redundancy. This technique isapplied prior to visual or digital interpretation of data [3].

Enhancement techniquesImage enhancement techniques improve the quality of an image as

perceived by a human. These techniques are most useful because manysatellite images when examined on a colour display give inadequateinformation for image interpretation. There exists a wide variety oftechniques for improving image quality. The contrast stretch, densityslicing, edge enhancement, and spatial filtering are the morecommonly used techniques. Image enhancement is attempted after theimage is corrected for geometric and radiometric distortions. Imageenhancement methods are applied separately to each band of amultispectral image. In the present study spatial filtering is used todelineate the lithology.

Spatial filteringA parameter of Remotely Sensed images is spatial frequency defined

as number of changes in Brightness Value per unit distance for anyparticular part of an image. In other words filtering is the process bywhich the tonal variations in an image in selected ranges orfrequencies of the pixel values, are enhanced or suppressed. Thisprocess selectively enhances or suppresses particular wavelengths orDN values with an image. If there are very few changes in BrightnessValue once a given area in an image, this is referred to as low frequencyarea. Conversely, if the brightness value changes dramatically overshort distances, this is an area of high frequency. Spatial filtering is theprocess of dividing the image into its constituent spatial frequencies,and selectively altering certain spatial frequencies to emphasize someimage features. This technique increases the analyst’s ability todiscriminate detail. The three types of spatial filters used in remotesensor data processing are Low pass filters, Band pass filters and Highpass filters. The mechanics of the spatial filtering involves themovement of the filter mask over the image. Convolution filter is mostcommonly used filters in image enhancement. In this method the filtermask is called convolution mask or convolution kernel. These kernelsare generally square in shape and are of odd number of pixels in sizelike 3 × 3, 5 × 5, 7 × 7 etc. The high pass and low pass convolutionfilters emphasize the high frequency and low frequency featuresrespectively.

Results and Discussion

DN Values of rocks of study areaDigtal Numbers (DN values) are intensity values representing the

solar radiance in a given wave length band reflected from the ground.In an image the actual intensity of reflected light can be derived fromthe pixel digital number. The DN values exerted various rock units inthe present study area are given in the Table 1 [14].

Principal component analysisThe multispectral image data is usually strongly correlated from one

band to the other. The level of a given picture element on one band canto some extent be predicted to some extent from the level of the samepixel in another band. Principal component analysis is a pre-processing transformation that creates new images from theuncorrelated values of different images.

Citation: Raghu BK, Keshava KKPL, Ramakrishna P (2017) Lithological Discrimination by Selective Image Processing Techniques. J RemoteSensing & GIS 6: 221. doi:10.4172/2469-4134.1000221

of 5


Volume 6 • Issue 4 • 1000221

Sl.No

Name of theRock

Band 1 Band 2 Band 3 Band 4

Min. Max. Mean Min. Max. Mean Min. Max. Mean Min. Max. Mean

1 PGC 61 149 94 46 156 95 42 150 90 45 128 89

2 GulcheruQuartzites 61 123 80 44 124 76 48 145 75 43 105 70

3 VempalliDolomite 64 141 87 46 149 90 49 140 86 47 113 84

4 VempalliLimestone 64 124 89 44 133 87 50 164 84 47 105 80

5 Tadipatri Shale 65 147 104 47 163 102 48 151 91 40 235 82

6 Volcanic Flow 65 132 91 52 131 85 47 135 73 48 122 74

Table 1: DN Values shown by various rock types of study area.

This is accomplished by a linear transformation of variables thatcorresponds to a rotation and translation of the original coordinatesystem. Principal component analysis operates on all bands together.In the present context, the study area composed several rock typesamong which quartzite, limestone, dolomite, shale and granite are lightcoloured and thus there is plausible correlation of bands of image data.PCA technique is applied to reduce the number of collelated band ofthe image to few uncorrelated bands and the visual interpretation forground features is highly enhanced. The enhanced image is effectivelyused as a base for ground truth collection during the classification ofvarious geological features of the study area (Figure 3).

Figure 3: Principal Component Analysis.

Low pass filtering in the spatial domainImage enhancements that de-emphasize or block the high spatial

frequency are low-frequency or low-pass filters. The simplest low-frequency filter evaluates a particular input pixel brightness value,BVin, are the pixels surrounding the input pixel, and outputs a newbrightness value, BVout, is the mean of this convolution. The size of theneighbourhood convolution mask or kernel (n) is usually 3 × 3, 5 × 5, 7× 7, or 9 × 9 (Figures 4-6).

Figure 4: 3 × 3 Low pass filtering.

The simple smoothing operation will, however, blur the image,especially at the edges of objects. Blurring becomes more severe as thesize of the kernel increases. Using a 3 × 3 kernel can result in the low-pass image being two lines and two columns smaller than the originalimage. Techniques that can be applied to deal with this probleminclude (1) artificially extending the original image beyond its borderby repeating the original border pixel brightness values or (2)


of 5


Volume 6 • Issue 4 • 1000221

replicating the averaged brightness values near the borders, based onthe image behaviour within a view pixel of the border. The mostcommonly used low pass filters are mean, median and mode filters.



High pass filtering in the spatial domainHigh-pass filtering is applied to imagery to remove the slowly

varying components and enhance the high-frequency local variations.

Brightness values tend to be highly correlated in a nine-elementwindow. Thus, the high frequency filtered image will have a relativelynarrow intensity histogram. This suggests that the output from mosthigh-frequency filtered images must be contrast stretched prior tovisual analysis (Figures 7-9).

Figure 7: 3 × 3 High pass filtering.



of 5


Volume 6 • Issue 4 • 1000221


ConclusionsDigital image processing’s of satellite data can be primarily grouped

into three categories: Image Rectification and Restoration,Enhancement and Information extraction. Image rectification is thepre-processing of satellite data for geometric and radiometricconnections. Enhancement is applied to image data in order toeffectively display data for subsequent visual interpretation.Information extraction is based on digital classification and is used forgenerating digital thematic map.

DN valuesThe study of DN values of the IRS P6 LISS III image reveals that

mean DN values between 89 and 95 denote granitic rocks, mean DNvalues between 70 and 80 indicate quartzites, mean DN values between84 and 90 corresponds to dolomites, mean DN values between 80 and89 denote limestones, mean DN values between 82 and 104 representshale and mean DN values between 73 and 91 indicates volcanic flows(Table 1).

PCA of image of study areaPrincipal Component Analysis of the image of the study area has

shown a clear variation between rocks like quartzites, limestone,

dolomite, shale and granite, composed of lighter minerals and alsothose of volcanic flows composed of darker minerals. Quartz andfeldspar rich granites and granodiorites have shown very lighter toneand mafic mineral bearing rocks such as dykes and volcanic flows haveshown very darker tone. Ferrugenous quartzites have shownmoderately darker tone.

Low pass filtering techniquesLow pass filtering technique of the order of 3 × 3, 5 × 5 and 7 × 7 is

applied to the image of the study area, the enhanced image has shownspecific demarcation between varied rock types (Figures 4-6). Darkertone has shown much darker and lighter appeared still lighter than theoriginal image. Litho units such as granites, granitoids, gneisses andschists of PGC, dolomites, limestones, shales and basic intrusivesappeared with sharp edges. More sharpened edges have been observedin 5 × 5 low pass filtering and 7 × 7 low pass filtering. Lineaments likefaults planes, dyke intrusions are seen more clearly.

High pass filtering techniquesHigh pass filtering techniques of the order 3 × 3, 5 × 5 and 7 × 7

when applied the image has shown much variations in relation to theLow pass filtering. In 3 × 3 High pass filtering, the image has shownhighly sharpened edges for PGC, quartzites, dolomites, limestones,shales and volcanic flows. Lineaments like faults and dykes are elevatedvery clearly (Figure 7). In 5 × 5 and 7 × 7 High pass filtering, rockscomposed of dark coloured minerals like mafic dykes intruded in thePGC and volcanic flows between Vempalli formation and Pulivendlaformation, ferruginous quartzites of Gulcheru formation appearedwith more darker tone, while other rocks like PGC, Vempalli dolomitesand Tadipatri shales appeared with very lighter tone.

The selective Digital Image Processing techniques like specialfiltering through low pass filtering and high pass filtering in the orderof 3 × 3, 5 × 5 and 7 × 7, PCA can be applied for the identification ofvarious rock types in a satellite image. This kind study is useful inmapping, mineral exploration and ground water explorations.

References1. Syed AA, Umair A (2015) Ligho-Structural mapping of Sind catchment

(Kashmir basin), N-W Himalaya, using Remote Sensing and GISTechniques. International Journal of Science and Research 4: 1325-1330.

2. NagarajaRao BK, Rajaurkar ST, Ramalingaswamy G (1987) Stratigraphy,Structure and Evolution of the Cuddapah Basin-Memoir. GeologicalSociety of India 6: 33-86.

3. Lillesand T, Kiefer RW, Chipman J (2012) Remote Sensing and ImageInterpretation. 6th edn, John Wiley & Sons, New Jersey, USA, p: 257.

4. Mohamed FS, Safaa MH (2012) Application of Remote Sensing inlithological discrimination and geological mapping of Precambrianbasement rocks in the Eastern Desert of Egypt. The 33rd AsianConference on Remote Sensing.


of 5


Volume 6 • Issue 4 • 1000221

e journal of remote sensing & gis - longdom · remote sensing by satellites and their study...

Documents